Southeast CASC Research Ecologist Mitch Eaton, along with USGS Researchers Julien Martin and Simeon Yurek and researchers at the University of South Florida and University of Florida, published SiteOpt: an Open-source R-package for Site Selection and Portfolio Optimization, an open-access article in journal Ecography on September 22, 2021.
The paper describes a new software package in R called SiteOpt, designed for spatial conservation planning and other spatial or non-spatial portfolio optimization problems, such as budget allocation. Systematic conservation planning (i.e., reserve design) is a critical tool in the global change adaptation arsenal. Protecting land is hugely expensive and such long-term investments are high-risk given climate change uncertainties. Responding to these challenges requires careful consideration of tradeoffs between cost, desired benefits, and risks from correlated uncertainty. The user-friendly R package, SiteOpt, produces globally optimal or near-optimal solutions for large-scale portfolio-type problems, and includes options for applying several powerful solver algorithms to handle large-dimension problems. The software considers multiple objectives, uncertainty, risk, and constraints (e.g., connectivity, budget) in identifying optimal spatial designs, resource allocations, species selection, and other portfolio decisions.
The authors provide examples to demonstrate the capability of SiteOpt in addressing large spatial conservation planning problems, including a case study (described by Eaton et al., 2019) that examined planning decisions for Cape Romain National Wildlife Refuge, a coastal wildlife refuge that currently faces impacts of sea-level rise and land-use change (see Modern Portfolio Theory Aids Reserve Design Under Climate and Land Use Change.)
SiteOpt can be installed via GitHub, https://github.com/paymanghasemi/SiteOpt. The software license, installation guidelines, user manual, examples and several supplementary short videos can all be found at the GitHub page.
Conservation planning involves identifying and selecting actions to best achieve objectives for managing natural, social and cultural resources. Conservation problems are often high dimensional when specified as combinatorial or portfolio problems and when multiple competing objectives are considered at varying spatial and temporal scales. Although analytical techniques such as modern portfolio theory (MPT) have been developed to address these complex problems, open source computational platforms for executing these approaches are not readily available. We present a user-friendly R-package called SiteOpt for optimization of binary decisions while explicitly considering environmental or economic uncertainty and the risk tolerance of decision makers. We illustrate the package with spatially-explicit site selection problems (i.e. spatial conservation planning), including an option for divestment (i.e. selling assets), when accounting for future uncertainties in designing conservation areas. The tool is applicable to both spatial and non-spatial problems, such as budget allocation or species selection. Constraints for spatial design and spatial dependencies (e.g. connectivity among sites) can also be specified in SiteOpt. Users can optimize site selection based on two competing objectives by solving for the Nash bargaining solution. Importantly, by quantifying uncertainty and asset spatial correlation, a measure of risk can be included as one such objective to be traded off against portfolio benefits. Thus, SiteOpt can be used to explicitly manage risk in portfolio-based spatial optimization. This tool facilitates decisions in a variety of problem settings, including reserve selection, invasive species management, allocation of law enforcement activities for conservation, budget allocation and asset selection under uncertainty and risk.
Ghasemi Saghand, P., Haider, Z., Charkhgard, H., Eaton, M., Martin, J., Yurek, S. and Udell, B.J. (2021), SiteOpt: an open-source R-package for site selection and portfolio optimization. Ecography. https://doi.org/10.1111/ecog.05717